Altera and Software Licensing

I want to distill some of the lessons learned in my dialogue with a couple of commenters over at Conglomerate recently, while writing about the Ninth Circuit’s recent decision in Altera v. Clear Logic. Altera looks like a relatively minor case in the grand scheme of intellectual property policy, but it teaches some noteworthy lessons about technology, licensing, IP rights, and competition.

Altera and Clear Logic are competitors in the integrated circuit business. Altera works at the high end, helping customers design programmable ICs (what it calls Complex Programmable Logic Devices, or CPLDs, and what are often generally called FPGAs, or Field Programmable Gate Arrays), and then selling the programmed ICs themselves. CL works at the low end, producing ASICs but also taking designs — including designs for programmable ICs worked out by Altera customers, who are using licensed Altera design software — and converting them into cheaper ASICs. Altera sued CL for infringement of Altera’s mask work rights under the Semiconductor Chip Protection Act (the rarely-litigated SCPA of 1984) and for interference with Altera’s licensing contracts with its customers. A jury returned a verdict for Altera; the Ninth Circuit affirmed.

Noteworthy dimension of the case number 1: As a result of the court’s decision, Altera has successfully created a software license under which it “owns” what gets produced when its licensees use the software. The Altera software generates a bitstream that is used to program Altera ICs. The Altera software license says that it can be used only to create Altera products, i.e., relatively expensive Altera ICs. As relevant to the lawsuit, CL’s business depends on taking these “Altera” bitstreams and converting them into instructions for lasers to follow in burning relatively cheap CL ASICs. The Ninth Circuit says that CL, and Altera’s customers, can’t do that.

Noteworthy dimension of the case number 2: From CL’s standpoint, and from the standpoint of Altera customers who use CL’s services, all of this sounds terribly onerous and coercive. Altera customers pay to use Altera services, get a product (bitstream), then shop that product around to a cheaper manufacturing source for production. The Ninth Circuit says that they can’t do that. But the coercion results at least in part from the nature of the technological advances that give us programmable ICs: design services and fabrication services can now be unbundled, technologically speaking, in new ways, and rebundled, contractually speaking, in other ways. We saw this problem before in what then appeared to be a fairly limited technological context: Could Microsoft lawfully declare that its Windows operating system software included a web browser? The DC Circuit said: for all practical purposes, if not for all formal legal purposes, yes. It turns out that the “what is the thing being sold (or licensed)” question is going to be pervasive across digital technologies.

Noteworthy dimension of the case number 3: Where the commenters were especially helpful was in teasing out the competitive implications of the case. Altera isn’t alone in the programmable IC business. Altera customers who want designs for FPGAs have options; customers who want ASICs instead also have options.

Some of the dimensions of that competitive environment include the location of the ASIC design and prototyping work, which may be done in the U.S. or which may be done overseas. The commenters argue that CL’s loss means that companies that want to design and prototype ASICs using customer bitstream data will move that business overseas to escape possible liability under extensions of the Ninth Circuit’s ruling.

A ruling for CL in this case not only would (allegedly) have limited that effect (though I still suspect that the economics of the IC business would push in that direction anyway), but it also would have pushed the industry in the direction of standards and then competition in the market for FPGA features within the bounds of those standards. The actual ruling in Altera’s favor allows Altera to lock-in its customers, forces FPGA competitors to rely on incompatible data formats, and raises costs for customers looking to use bitstream data to prototype cheap ASICs. It also helps Altera maintain its competitive position in a market that it does not control and protect its incentive to develop its own CPLD technology. It also means that the technological and contractual flexibility offered by the development of FPGA technology as an alternative to ASIC development isn’t frozen prematurely, or at all, by invocation of an IPR statute — the SCPA — that was designed for a much older technological era.